replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[OMPT]: Fix the order of implicit_task_end_events 

For implcit barriers in simple parallel for loops, the order of the OMPT events
was wrong.  The barrier_{begin,end} events came after the implcit_task_end
event for the implcit barrier at the end of the parallel region. This is wrong
because the implicit task executes the barrier before ending. This patch fixes
the order of the event: It will be triggerd now just before
__kmp_pop_current_task_from_thread() is called.

Patch by Tim Cramer

Differential Revision: http://reviews.llvm.org/D16347

llvm-svn: 258866
This commit is contained in:
Jonathan Peyton 2016-01-26 21:45:21 +00:00
parent 00adc1e105
commit b4c73d8d8a
1 changed files with 31 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
openmp/runtime/src/kmp_runtime.c
View File

@ -2475,6 +2475,18 @@ __kmp_join_call(ident_t *loc, int gtid
}
KMP_DEBUG_ASSERT( root->r.r_in_parallel >= 0 );
#if OMPT_SUPPORT && OMPT_TRACE
if(ompt_enabled){
ompt_task_info_t *task_info = __ompt_get_taskinfo(0);
if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
parallel_id, task_info->task_id);
}
task_info->frame.exit_runtime_frame = 0;
task_info->task_id = 0;
}
#endif
KF_TRACE( 10, ("__kmp_join_call1: T#%d, this_thread=%p team=%p\n",
0, master_th, team ) );
__kmp_pop_current_task_from_thread( master_th );
@ -5504,6 +5516,12 @@ __kmp_launch_thread( kmp_info_t *this_thr )
/* have we been allocated? */
if ( TCR_SYNC_PTR(*pteam) && !TCR_4(__kmp_global.g.g_done) ) {
#if OMPT_SUPPORT
ompt_task_info_t *task_info;
if (ompt_enabled) {
task_info = __ompt_get_taskinfo(0);
}
#endif
/* we were just woken up, so run our new task */
if ( TCR_SYNC_PTR((*pteam)->t.t_pkfn) != NULL ) {
int rc;
@ -5517,8 +5535,7 @@ __kmp_launch_thread( kmp_info_t *this_thr )
this_thr->th.ompt_thread_info.state = ompt_state_work_parallel;
// Initialize OMPT task id for implicit task.
int tid = __kmp_tid_from_gtid(gtid);
(*pteam)->t.t_implicit_task_taskdata[tid].ompt_task_info.task_id =
__ompt_task_id_new(tid);
task_info->task_id = __ompt_task_id_new(tid);
}
#endif
@ -5533,8 +5550,7 @@ __kmp_launch_thread( kmp_info_t *this_thr )
#if OMPT_SUPPORT
if (ompt_enabled) {
/* no frame set while outside task */
int tid = __kmp_tid_from_gtid(gtid);
(*pteam)->t.t_implicit_task_taskdata[tid].ompt_task_info.frame.exit_runtime_frame = 0;
task_info->frame.exit_runtime_frame = 0;
this_thr->th.ompt_thread_info.state = ompt_state_overhead;
}
@ -5545,6 +5561,17 @@ __kmp_launch_thread( kmp_info_t *this_thr )
}
/* join barrier after parallel region */
__kmp_join_barrier( gtid );
#if OMPT_SUPPORT && OMPT_TRACE
if (ompt_enabled) {
if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
int my_parallel_id = (*pteam)->t.ompt_team_info.parallel_id;
ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
my_parallel_id, task_info->task_id);
}
task_info->frame.exit_runtime_frame = 0;
task_info->task_id = 0;
}
#endif
}
}
TCR_SYNC_PTR((intptr_t)__kmp_global.g.g_done);
@ -6858,17 +6885,6 @@ __kmp_invoke_task_func( int gtid )
);
}
#if OMPT_SUPPORT && OMPT_TRACE
if (ompt_enabled) {
if (ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)) {
ompt_callbacks.ompt_callback(ompt_event_implicit_task_end)(
my_parallel_id, my_task_id);
}
// the implicit task is not dead yet, so we can't clear its task id here
team->t.t_implicit_task_taskdata[tid].ompt_task_info.frame.exit_runtime_frame = 0;
}
#endif
#if USE_ITT_BUILD
if ( __itt_stack_caller_create_ptr ) {
__kmp_itt_stack_callee_leave( (__itt_caller)team->t.t_stack_id ); // inform ittnotify about leaving user's code